1. Field of the Invention
This invention is related to noise reduction in communication systems such as Ethernet. More particularly, this invention is related to reduction of D.C. noise components in a signal received from a communication channel.
2. Background Information
The structures of digital communications such as Ethernet, as described in the Institute of Electrical and Electronic Engineers (IEEE) specification IEEE 802.3, are well known in the art. To improve reception of encoded signals from a communication channel, noise components must be determined and subtracted from the received signal. One of the noise components induced within the communication channel is referred to as D.C. noise. Refer to FIGS. 2a and 2b to understand the definition and the effect of D.C. noise components on the received signal ai. FIG. 2a shows an eye diagram of a received signal ai of a five-level amplitude-modulated signal. Each voltage level represents one of the code values of the digital signal being transferred. In an ideal communication system, each “eye” of the eye diagram would be nearly rectangular. The more “closed” the “eye,” the more noise induced on the communication channel. Low frequency or D.C. noise components that are a result of the encoding of the digital data or that are induced on the communication channel may actually shift the D.C. voltage level as shown in FIG. 2b. In this example, the whole signal ai is shifted by a D.C. voltage level Δdc. The effect of the D.C. noise components is to degrade the performance of the circuitry that decoded the received signal ai to recover the digital data. This implies a higher error rate of the data received, thus requiring more robust error recovery and slower data transmission.
Further, as can be seen in FIG. 2b, the receiver must have a wider dynamic range. That is the voltage range that the receiver recognizes, as the signal ai must be larger.
Refer now to FIG. 1 for a discussion of a receiver and a D.C. noise cancellation circuit of the prior art. The multiple level amplitude-modulated signal ai is transferred to the communication channel. The received signal airec having D.C. noise components, as shown in FIG. 2b, is the input to the receiver. Typically, it is desirable to remove as much of the D.C. noise components as early as possible in the receiving of the signal airec. Typically, a D.C. noise canceling signal is subtractively combined in a first summing circuit Σ1 with the received signal airec to remove the D.C. noise components. The signal ainc with the D.C. noise components removed is the input to the analog-to-digital converter.
The analog-to-digital converter creates a set of sampled digital data hk indicating the amplitude of the received signal ainc with the D.C. noise components removed. The sampled digital data is created at discrete periods of a sampling clock and is retained or buffered as needed in a memory or registers (not shown).
As is known in the art, the communication channel acts as a lowpass filter causing what is termed intersymbol interference where noise components from adjacent symbols of the encoded data interfere with the current symbol. The feedforward or feedback equalizer removes any of the intersymbol interference to create the equalized sampled digital data of the received signal fk. The equalized sampled digital data fk is the input of the decision circuit that determines an estimate of the transmitted value of the signal âk.
An error signal εk is determined as the difference between the equalized sampled digital data fk and the estimated value of the signal âk. The equalized sampled digital data fk and the estimated values of the signal âk are the inputs to the second summing circuit Σ2. The second summing circuit Σ2 subtractively combines the equalized sampled digital data fk and the estimated values of the signal âk to form the error signal εk.
The error signal εk is used to determine the level of the D.C. noise component that needs to be removed from the received signal.
The error signal εk is the input to the D.C. noise canceling circuit DCC. The error signal, further, is the input of the multiplier circuit M1. The second input of the multiplier circuit M1 is a gain constant μdc. The gain constant μdc is chosen to be sufficiently small to make the noise cancellation stable, but sufficiently large to track any slow variation in the D.C. voltage level of the received signal airec. The range of the gain constant μdc is dependent on the condition of the communication channel.
The output of the multiplier circuit M1 is one input of a third summing circuit Σ3. The second input of the third summing circuit Σ3 is the value of the D.C. noise cancellation signal for the previous sample period, and the output of the third summing circuit Σ3 is the D.C. noise cancellation signal for the previous sampling period of the received signal airec. The output of the third summing circuit Σ3 is the input of the delaying circuit τd1. The output of the delaying circuit τd1 is the input to the first summing circuit Σ1 to remove the D.C. noise component from the received signal airec.
U.S. Pat. No. 4,486,740 (Seidel) describes a circuit for DC noise cancellation in ternary-coded data systems. An encoder processes an input signal to produce a ternary coded data stream having suppressed DC. The data stream is augmented with a compensating set of code symbols as determined by the number of positive and negative code symbols in the data stream as well as all prior compensating code symbols. A decoder processes the received signal to extract the symbols in the data stream corresponding to the input signal. In order to achieve a pre-selected end-to-end transmission rate with the encoder-decoder combination, the rate of the signal propagated between encoder and decoder is increased to compensate for the appended code symbols.